[10p-N304-1~5] 18.4 Methods, Theory, Foundations, Others

2:15 PM - 2:30 PM JST(5:15 AM - 5:30 AM UTC)

[10p-N304-4] Expanding and Deepening Autonomous Exploration of Physical Property Information from Materials Experimental Databases Using Multi-Agent AI

〇Yusuke Hashimoto¹, Takaaki Tomai¹ (1.Tohoku Univ.)

Keywords:

Agent AI,Thermoelectric materials,Autonomous Exploration

To elucidate the process of knowledge exploration and accumulation by multi-agent AI, we conducted experiments in which the number of themes and cycles was systematically varied. Each project inherits knowledge obtained from previous projects and repeatedly extracts knowledge while revising its hypotheses. Knowledge maps were then constructed from the resulting knowledge bases and compared. The results suggest that this system has the potential to effectively explore, expand, and deepen physical property knowledge embedded in materials experimental databases.